Emotion regulation as mediator of treatment outcome in therapy for deliberate self-harm.

(1)

deliberate self-harm.

Slee, N.; Spinhoven, P.; Garnefski, N.; Arensman, E.

Citation

Slee, N., Spinhoven, P., Garnefski, N., & Arensman, E. (2008). Emotion regulation as mediator of treatment outcome in therapy for deliberate self-harm. Clinical Psychology And

Psychotherapy, 15, 205-216. Retrieved from https://hdl.handle.net/1887/14271

Version: Not Applicable (or Unknown)

License: Leiden University Non-exclusive license Downloaded from: https://hdl.handle.net/1887/14271

Note: To cite this publication please use the final published version (if applicable).

(2)

Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/cpp.577

Emotion Regulation as Mediator of Treatment Outcome in Therapy for Deliberate Self-Harm

Nadja Slee,¹* Philip Spinhoven,² Nadia Garnefski¹ and Ella Arensman³

1 Unit of Clinical Psychology, Institute of Psychological Research, Leiden University, Wassenaarseweg, Leiden, the Netherlands

2 Unit of Clinical Psychology, and the Department of Psychiatry of the Leiden University Medical Centre, Institute of Psychological Research, Leiden University, Wassenaarseweg, Leiden, the Netherlands

3 National Suicide Research Foundation, Cork, Ireland

This study presents the outcomes of mediator analyses as part of a randomized controlled trial of Cognitive-Behavioural Therapy (CBT) for young people who engage in deliberate self-harm (DSH). The study involved 90 people, aged 15–35 years, who were randomly assigned to CBT in addition to treatment as usual or to treatment as usual only. The fi ndings showed that changes in DSH were partially mediated by changes in emotion-regulation diffi culties, particularly diffi culties with impulse control and goal-directed behaviours. In addition, the potential mediating role of symptoms of depression, anxiety and suicidal cognitions was examined. Although the CBT intervention signifi cantly reduced depression, anxiety and suicidal cognitions, these measures of symptom severity did not play a medi- ating role. These fi ndings suggest that interventions for DSH should not primarily focus on mental disorders associated with DSH, but should be DSH-specifi c and should target specifi c emotion-regulation diffi culties. Copyright © 2008 John Wiley & Sons, Ltd.

* Correspondence to: Nadja Slee, Department of Clinical, Health and Neuropsychology, Wassenaarseweg 52, P.O. Box 9555, 2300 RB Leiden, The Netherlands.

E-mail: nadja.slee@fsw.leidenuniv.nl

provided evidence consistent with the emotional dysregulation hypothesis for DSH. First, descrip- tive studies consistently show that compared with psychiatric and non-psychiatric control participants, DSH patients (often diagnosed with borderline personality disorder) score signiﬁ cantly higher on self-report measures tapping emotion regulation difﬁ culties, including lower emotional awareness and clarity (Leible & Snell, 2004; Levine, Marziali,

& Hood, 1997; Slee, Garnefski, Van der Leeden, Arensman, & Spinhoven, 2008a), lower acceptance of emotions (Slee, Garnefski, Spinhoven, & Arens- man, 2008b) and difﬁ culty controlling behaviour when experiencing negative emotions (Leible &

Snell, 2004; Slee et al., 2008b; Yen, Zlotnick, &

Costello, 2002). Furthermore, these patients make greater use of avoidant emotion regulation strategies (Bijttebier & Vertommen, 1999). A limitation is that these studies rely on self-report measures INTRODUCTION

Emotion regulation has a central role in theories of deliberate self-harm (DSH) (Linehan, 1993). The conceptualization of emotion regulation used here (see Gratz & Roemer, 2004) emphasizes the func- tional nature of emotional responses, with emotion dysregulation referring to maladaptive responses to emotions. Speciﬁ cally, Gratz and Roemer (2004) broadly deﬁ ne emotion regulation as the awareness, understanding and acceptance of emotions, as well as the ability to control behaviour in the context of emotional distress. Several lines of research have

(3)

only, which may have resulted in underreport- ing or over-reporting of emotion regulation diffi culties. Second, an experimental investigation of distress tolerance among patients with borderline personality disorder revealed an unwillingness to experience unpleasant emotions (Gratz, Rosenthal, Tull, Lejuez, & Gunderson, 2006), which is consistent with a recent theoretical model that describes the primary function of DSH as the avoidance of unpleasant emotions (Chapman, Gratz, & Brown, 2006). However, this study involved a small and homogenous sample of participants without mood disorders, limiting the generalizability of the results. Further support for the emotion regula- tory function of DSH comes from research among female members of a Dutch DSH support organization. This study shows that negative emotions are highest immediately before the episode of DSH, drop markedly after and increase again one day after the episode of DSH, while the reversed pattern is observed for positive emotions (Kam- phuis, Ruyling, & Reijntjes, 2007). Although these fi ndings are consistent with the emotional dysregulation hypothesis for DSH, the study sample is a self-selected group of support-seeking females with DSH and borderline personality disorder. In addition, the study draws on introspection and retrospective self-report. Finally, during a 24-hour naturalistic psychophysiological ambulatory monitoring approach, it was found that patients diagnosed with borderline personality disorder reported more negative emotions, fewer positive emotions and a greater intensity of negative emotions than healthy controls (Ebner-Priemer et al., 2007a). In a second study it was found that these patients, on average, took little me to fl uctuate from a positive mood to a negative mood, which suggests affective instability (Ebner-Priemer et al., 2007b). Affective instability is the feature of borderline personality disorder most strongly related to DSH. However, Stone, Broderick, Shiffman, and Schwartz (2004) argue that the retrospective assessment of mood fl uctuation, as demonstrated in the study of Ebner- Priemer et al. (2007b), is problematic. Furthermore, the ambulatory monitoring studies only assess emotional experience and do not ask participants to report daily life stressors during the monitoring.

As a result, we do not know if the participants exhibit heightened sensitivity to discrete stimuli and respond more intensively to events than to controls (as would follow from Linehan’s theory of emotional dysregulation).

Cognitive behavioural therapy (CBT) for DSH is based on a theoretical model which assumes that

vulnerability to DSH can be changed by changing emotion regulation deﬁ cits (Linehan, 1993; Slee, Arensman, Garnefski, & Spinhoven, 2007). Treat- ment is directed at developing emotion regulation skills for coping with situations that trigger DSH, and modifying cognitions and behaviour that interfere with effective emotion regulation.

Randomized controlled trials (RCT’s) of CBT for DSH are limited and their results are inconsistent.

An RCT of brief CBT versus usual care showed that brief CBT was of limited effi cacy in reducing DSH (Tyrer et al., 2003), whereas an RCT of cognitive therapy (CT) reported favourable outcomes with regard to the number of suicide attempts, depression severity and thoughts of hopelessness (Brown et al., 2005). The BOSCOT trial evaluated the clinical effectiveness and cost-effectiveness of CBT for suicidal patients with borderline personality disorder (Davidson et al., 2006; Palmer et al., 2006). Compared with patients in the TAU group, patients assigned to CBT were less likely to attempt suicide (Davidson et al., 2006). A RCT of schema-focused therapy among patients with borderline personality disorder also reported positive outcomes with regard to the number of episodes of DSH (Giesen-Bloo et al., 2006). To date, dialectical behavioural therapy is the only CBT intervention studied in more than one RCT, and has consistently been found to reduce self-injury in (female) patients with borderline personality disorder (Linehan et al., 2006). Furthermore, a recent RCT showed that DSH patients who received CBT in addition to usual care had signifi cantly greater reductions in DSH, suicidal cognitions, symptoms of depression and anxiety and signifi cantly greater improvements in self-esteem and problem-solving ability than controls (Slee et al., 2008a).

However, as the evidence supporting the efﬁ cacy of CBT for DSH accumulates, there are no studies investigating the mechanism of action of CBT.

Understanding the mechanisms through which CBT operates would advance the development of innovative treatment strategies. Identifying mechanisms of change promises not only to improve treatment but also to enhance understanding of the nature of DSH. If a treatment has its effects by inﬂ uencing a particular process, this ﬁ nding points to the importance of this process in the maintenance of the disorder. Analysing these mechanisms therefore also provides an indirect test of the theoretical model of mechanisms maintaining the disorder.

On the basis of contemporary psychological theories of DSH, it makes sense to ask whether

(4)

the changes resulting from CBT are mediated by changes in emotion-regulation difﬁ culties.

The main objective of the present study was to investigate whether changes in speciﬁ c emotion- regulation difﬁ culties in DSH patients treated with CBT indeed mediated treatment outcome.

Considering the targets of CBT, it was expected that diffi culties with impulse control, diffi culties with goal-directed behaviours, lack of awareness of emotions and non-acceptance of emotions would be the most important mediators of treatment outcome. Few empirical studies have exam- ined the association between particular aspects of emotion regulation diffi culties and DSH. One study found that none of the correlations between DSH and aspects of emotion regulation differed signifi cantly from one another (Gratz & Roemer, 2004). However, this study involved a non-clinical population of college students. In another study, a clinical group of young women who harmed themselves was compared with a group of young women without a history of DSH across various aspects of emotion regulation. Controlling for depression severity, non-acceptance of emotions independently predicted DSH (Slee et al., 2008b).

Given the sparse empirical evidence on particular emotion-regulation difﬁ culties and DSH, our expectations of mediation are largely based on the targets of CBT. We expected four aspects of emotion regulation to mediate treatment change:

diffi culties with impulse control, diffi culties engaging in goal-directed behaviours, lack of awareness of emotions and non-acceptance of emotions. In addition to these specifi c hypotheses, we explored the possible mediating role of lack of clarity of emotions and of limited access to emotion- regulation strategies. Finally, we investigated the possible mediating role of symptom severity (depression, anxiety and suicidal cognitions).

The hypothesis that the effects of CBT on DSH would be mediated by changes in speciﬁ c emotion- regulation difﬁ culties was tested in accordance with the approach advocated by Baron and Kenny (1986). According to Baron and Kenny, a variable may be called a mediator ‘to the extent that it accounts for the relation between the predictor and the criterion’ (Baron & Kenny, 1986, pp. 1176).

Figure 1 illustrates the Baron and Kenny approach.

The ﬁ rst model shows the direct effect of the treatment condition on DSH (C). The second model represents mediation: C′ is the direct effect of the treatment condition on DSH after controlling for the mediator. The results reported here are from a clinical trial that examined the outcome of CBT on

DSH (Slee et al., 2008a). We perform the mediator analyses using the baseline and 9-month follow- up data from this trial in order to test whether change in speciﬁ c emotion regulation difﬁ culties predicts reduction of DSH at 9-months follow-up in those who received CBT in addition to treatment as usual (TAU) compared with TAU only.

METHOD

Patients and Procedure

In order to be eligible, patients had to meet the following criteria: having recently engaged in DSH and being aged 15–35 years old. Patients were excluded if they reported severe psychiatric disorders requiring intensive inpatient treatment, were unable to converse in Dutch, had cognitive impair- ments or lived outside the region of Leiden.

Participants were recruited from March 2003 through April 2006 at the Leiden University Medical Centre and the local Mental Health Centre Rivierduinen. Those who agreed to participate and who were found to be eligible for the study (n = 90) were randomly assigned to 12 sessions of CBT in addition to TAU (n = 48), or to TAU only (n = 42). These participants were invited for subsequent assessments 3, 6 and 9 months following the baseline interview. For the present article, the baseline and 9-month follow-up measurements were used.

Our analyses were based on the intent-to-treat sample (n = 90). At baseline, there were no missing data. At 9-months follow-up, data were missing for 17 patients, all dropouts. Of the 17 dropouts, eight had dropped out right after the baseline interview and nine patients had dropped out during the course of the study. For each individual, missing values were replaced by the last observed value of that variable (using the so-called Last Observation Carried

Condition DSHΔ

MediatorΔ

DSHΔ Condition

C'

A B

C

Figure 1. Model of mediated intervention effects.

Dotted lines indicate that the mediator variable was controlled for. ∆ = change

(5)

Forward (LOCF) method). To control for the impact of missing data and the use of the LOCF method on the mediation analyses, we performed the analy- ses both in the intent-to-treat sample (n = 90) and in the completers sample (n = 73). The completers sample consisted of the sample of 90 patients minus the eight early dropouts and the nine patients who dropped out during the course of the study. A com- plete description of the ﬂ ow of participants through the study is given elsewhere (Slee et al., 2008a).

Measures Demographics

During a structured clinical interview, demographic information was obtained regarding age, gender, living situation, marital status, educational level, job status and nationality.

Outcome: DSH

The primary outcome measure of the study was the number of episodes of DSH in the past 3 months, which was assessed using a structured clinical interview. DSH was defi ned as including both deliberate self-poisoning (overdose) and deliberate self-injury (Hawton, Zahl, & Weatherall, 2003). An overdose was defi ned as the deliberate ingestion of more than the prescribed or recom- mended amount of chemical substances with the intention of self-harm. Patients were also asked about incidents of deliberate self-injury, which was defi ned as intentional self-injury, irrespective of the apparent purpose of the act, and included cutting, scratching, punching, kicking and head- banging. In this defi nition, both the original para- suicide defi nition of the WHO/Euro study and their current nomenclature of fatal and non-fatal suicidal behaviour is included, as well as habitual behaviours and self-injuries with no intent to die, which they exclude (De Leo, Bille-Brahe, Kerkhof,

& Schmidtke, 2004; Schmidtke, Bille-Brache, Leo, &

Kerkhof, 2004). Thus, all behaviour that was self- initiated with the intent to harm the body (regard- less of intent to die) is included. The study does not distinguish between suicidal acts and self-injury.

Emotion Regulation Difﬁ culties as Proposed Mediators of Treatment Outcome

The Difﬁ culties in Emotion Regulation Question- naire (DERS) (Gratz & Roemer, 2004) contains six dimensions of emotion regulation wherein difﬁ culties may occur, including (a) lack of awareness of emotional responses (e.g., ‘I pay attention to how

I feel’ = reverse-scored item), (b) lack of clarity of emotional responses (e.g., ‘I have diffi culty making sense out of my feelings’), (c) non-acceptance of emotional responses (e.g., ‘When I’m upset, I feel ashamed with myself for feeling this way’), (d) limited access to emotion regulation strategies perceived as effective (e.g., ‘When I’m upset, I believe that there is nothing I can do to make myself feel better’), (e) diffi culties controlling impulses when experiencing negative emotions (e.g., ‘When I’m upset, I feel out of control’), and (f) diffi culties engaging in goal-directed behaviours when experiencing negative emotions (e.g., ‘When I’m upset, I have diffi culty concentrating’). All questions are self-rated from 1 (almost) never to 5 (almost) always. Scores on the subscales range from 5–25 (‘Clarity’, ‘Goals’), from 6–30 (‘Awareness’, ‘Non- acceptance’) and from 7–35 (‘Impulses’, ‘Strate- gies’). All of the DERS subscales have adequate internal consistency, with alpha reliabilities of 80 or higher for each subscale (Gratz & Roemer, 2004). In this study we also found alpha reliabilities of 0.80 or higher for each subscale: 0.82 for lack of awareness, 0.87 for lack of clarity, 0.83 for non-acceptance, 0.85 for limited strategies, 0.90 for diffi culties controlling impulses, and 0.80 for diffi culties goals.

Symptom Severity (Depression, Anxiety and Suicidal Cognitions) as Proposed Mediators of Treatment Outcome

Depression was measured by the Beck Depres- sion Inventory II (BDI-II) (Beck, Steer, & Brown, 1996a): a 21-question depression scale with each answer rated 0–3. Scores range from 0–63. The test has high internal consistency with an alpha reliability of 0.91 (Beck et al., 1996a; Beck, Steer, Ball,

& Ranieri, 1996b). In this study we found an alpha reliability of 0.93.

Anxiety was measured by the Symptom Check- list-90 (Arindell & Ettema, 1986; Derogatis, Lipman,

& Covi, 1973), which is a self-report clinical rating scale of psychiatric symptomatology. It consists of 90 items in total, with 10 items for the subscale anxiety, ﬁ ve-point Likert, ranging from ‘not at all distressing’ (0) to ‘extremely distressing’ (4). Scores of this subscale range from 0–40. Previous studies have reported alpha-coefﬁ cients ranging from 0.71 to 0.91 for the anxiety subscale. In addition, test–retest reliabilities were found to be good and the subscale showed strong convergent validity with other conceptually related scales (Arrindell

& Ettema, 1986). In this study we found an alpha reliability of 0.93 for the anxiety subscale.

(6)

Suicidal cognitions were measured by the Suicide Cognition Scale (Rudd, Joiner, & Rajab, 2001): 20 questions about core beliefs of perceived burdensomeness (‘I am a burden to my family’), helplessness (‘No one can help solve my problems’), unlovability (‘I am completely unworthy of love’) and poor distress tolerance (‘When I get this upset, it is unbearable’), with each answer rated 1 (strongly disagree) to 5 (strongly agree). Scores on the total scale range from 20–100. Scores on the subscale Perceived Burdensomeness (two items) range from 2–10, scores on the subscale Helpless- ness (ﬁ ve items) range from 5–25, scores on the subscale Unlovability (six items) range from 6–30 and scores on the subscale Poor Distress Tolerance (seven items) range from 7–35. In this study we found an alpha reliability of 0.96 for the total scale, of 0.74 for Perceived Burdensomeness, of 0.88 for Helplessness, of 0.90 for Poor Distress Tolerance and of 0.89 for Unlovability.

Treatment Interventions

Cognitive–Behavioural Therapy Intervention (CBT-condition)

In addition to usual care (e.g., psychotropic medication, psychotherapy or psychiatric hospitalizations), participants in the CBT condition were to receive 12 outpatient CBT sessions specifi cally developed for preventing repeated DSH. Ten out of 12 sessions were given weekly; the last two sessions were follow-up sessions. Altogether, the intervention lasted 5.5 months. The central feature of this intervention was the identifi cation and modifi cation of the mechanisms that maintained the individual’s self-harming behaviour. Thus, the treatment started with the assessment of the most recent episode of DSH. The therapist then specifi ed how emotional, cognitive and behavioural factors played a role in the maintenance of DSH. Specifi c maintenance factors that were addressed included dysfunctional cognitions, emotion regulation diffi culties and poor problem solving. Changing emotion regulation diffi culties involved interventions geared towards mindfulness, acceptance and exposure with response prevention. Towards the end of therapy, relapse prevention was addressed as well. A manual was written to standardize the intervention (available on request).

All therapists were experienced practitioners of CBT and were experienced in working with patients who engage in DSH. Prior to taking part in the research project, they received two days of

training in the standardized protocol. At monthly meetings, the treatment sessions were reviewed and therapists could share their experiences with their colleagues.

Treatment as Usual (TAU) Comparison Condition For ethical reasons, participants in this group were free to pursue any treatment they deemed warranted. Most of the interventions involved a limited number (between 2 and 30) of sessions of individual psychotherapy such as CBT and inter- personal psychotherapy. Social skills training was also common, especially among adolescents and young adults. No treatment specifi c to DSH was reported. These treatments focused on specifi c psychiatric problems (e.g., depression) or on other needs of the patient (e.g., problems with housing, fi nances, social isolation). In addition to psycho- social interventions, the majority of the patients received psychotropic medication. The number of psychiatric hospitalizations was also recorded.

Data Analyses

We examined baseline differences between patients in CBT and TAU in demographic characteristics, DSH, symptom severity, and emotion-regulation difﬁ culties using t-test analyses for continuous variables and chi-square tests for categorical variables. In addition, we examined differences in DSH, symptom severity and emotion regulation during the 9-month follow-up between patients in CBT and TAU, using t-tests. Pearson correlations were calculated to examine the relationships between the subscales of emotion regulation difﬁ culties and DSH, as well as symptom severity and DSH.

The hypothesis that the effects of CBT on DSH would be mediated by changes in specifi c emotion regulation diffi culties was tested in accordance with approach advocated by Baron and Kenny (1986). Baron and Kenny (1986) proposed a four- step approach in which several regression analyses are conducted. First, treatment condition should predict change in the outcome (Path C). Second, treatment condition should predict change in the proposed mediator (Path A). Third, change in the mediator should be signifi cantly associated with change in the outcome in the CBT condition (Path B). Fourth, the effect of treatment condition on change in the outcome should be attenuated when change in the mediator is statistically controlled (Path C′) (see Figure 1). Separate linear regression analyses were used to examine mediation. In these

(7)

analyses, corrections were made for the baseline scores of DSH and the proposed mediators by cal- culating residualized change scores. Residualized change scores are the 9-month follow-up scores after statistically correcting for any baseline differences on this measure using linear regression analysis.

RESULTS

Group Comparisons at Baseline and 9-Month Follow-up

Of the 90 patients, 93% were female. Their mean age was 24.2 years (standard deviation [SD] = 5.6), 92% were of Dutch nationality, 71% lived alone and 74% were unmarried. In addition, 29% went to school or studied, 24% had a full-time or part- time job and 38% lived on social welfare beneﬁ ts.

There were no signiﬁ cant differences between the CBT condition and TAU condition with regard to these demographics.

Table 1 contains the means and SDs for DSH, symptom severity and emotion regulation diffi culties in both groups at baseline and 9-month follow-up. t-Tests showed no signifi cant baseline differences on DSH, measures of symptom severity and measures of emotion regulation diffi culties, except for lack of awareness of emotions, which

was signifi cantly more often reported in CBT than in TAU (t = 2.99, degree of freedom [df] = 88, p = 0.004) (see Table 1). The baseline difference in lack of awareness of emotions was controlled for at the 9-month follow-up. Furthermore, at 9- month follow-up, signifi cant group differences were found for all study variables, with patients in CBT reporting signifi cantly lower scores of DSH, measures of emotion regulation diffi culties and measures of symptom severity (see Table 1).

Pearson Correlations between DSH and Emotion-Regulation Diffi culties at Baseline Correlations between subscales ranged between 0.01 (lack of clarity and DSH) and 0.72 (diffi culties with impulse control and diffi culties with goal- directed behaviours) (see Table 2). Contrary to our expectations, positive correlations were found for DSH and only half of the DERS-subscales: lack of awareness, limited access to strategies and diffi culties with goal-directed behaviours.

Pearson Correlations between DSH and Symptom-Severity Measures at Baseline

Correlations between subscales ranged between 0.19 (DSH and suicidal cognitions) and 0.74 (depression and suicidal cognitions) (see Table 3).

Table 1. Means and standard deviations (SDs) at baseline and 9-month follow-up for the treatment outcome and proposed mediators

CBT (n = 42) TAU (n = 48) CBT (n = 42) TAU (n = 48)

Baseline 9 month follow-up

Variable Mean (SD) Mean (SD) Mean (SD) Mean (SD)

Outcome

DSH 14.63 (10.49) 11.32 (11.34) 2.06 (6.04) 5.39 (8.70)*

Proposed mediators: emotion regulation difﬁ culties (DERS subscales)

Lack of awareness of emotions 21.39 (4.60) 18.34 (5.07)* 16.67 (5.91) 19.12 (4.33)*

Lack of clarity of emotions 17.59 (4.02) 16.71 (5.24) 12.98 (5.21) 16.20 (4.58)*

Non-acceptance of emotions 20.92 (5.04) 20.88 (5.18) 14.37 (6.17) 19.49 (6.17)**

Limited strategies 22.63 (5.60) 23.59 (6.03) 15.98 (6.85) 22.78 (7.50)**

Difﬁ culties impulse control 25.45 (6.11) 26.56 (6.80) 17.98 (8.51) 25.41 (8.17)**

Difﬁ culties goal-directed behaviours 19.41 (3.54) 19.76 (3.65) 14.82 (5.52 19.02 (4.63)**

Proposed mediators: symptom severity

Depression (BDI-II) 32.53 (16.13) 34.24 (13.97) 14.33 (13.70) 30.12 (16.27)**

Anxiety (SCL-90) 29.84 (8.51) 28.27 (10.55) 20.65 (7.94) 28.02 (10.57)**

Suicidal cognitions (SCS) 58.61 (14.12) 62.90 (19.56) 40.29 (18.61) 56.76 (18.98)**

* p < 0.05; ** p < 0.01; *** p < 0.001.

DSH = deliberate self-harm. CBT = cognitive behavioural therapy. TAU = therapy as usual. BDI-II = Beck Depression Inventory II.

DERS = Difﬁ culties in Emotion Regulation Questionnaire.

SCL-90 = Symptom Checklist 90.

(8)

Pearson Correlations between Emotion Regulation Variables and DSH

at 9-Month Follow-up

Correlations between subscales ranged between 0.13 (lack of clarity and DSH) and 0.86 (difﬁ culties with impulse control and difﬁ culties with goal- directed behaviours) (see Table 4). Positive correlations were found between DSH and all of the DERS subscales, except for lack of clarity.

Effects of Treatment on Treatment Outcome (DSH) (Mediation Test—Step 1)

To test whether treatment condition predicted change in outcome (Path C), we used a linear regression analysis with treatment condition as the predictor variable (coded 0 for TAU and 1 for CBT) and the residual change score of DSH (which is the DSH score at 9-month follow-up corrected for any baseline differences on this measure at baseline) as the outcome variable. Treatment condition predicted change in DSH at 9-month assessment (standardized β = −0.298, p = 0.004). Thus, the ﬁ rst requirement for mediation was fulﬁ lled.

Effects of Treatment on the Proposed Mediators (Mediation Test—Step 2)

To determine whether treatment condition (coded 0 for TAU and 1 for CBT) predicted change in the mediators (Path A), we used separate regression analyses with condition as the predictor variable and the residualized change score of each of the proposed mediators (emotion regulation difﬁ culties and symptom severity) as the outcome variable.

Treatment condition predicted change in emotion regulation difﬁ culties at 9-month assessment: lack of awareness (standardized β = −0.383, p < 0.001), lack of clarity (standardized β = −0.438, p < 0.001), non-acceptance of emotions (standardized β =

−0.398, p < 0.001), limited strategies (standardized β

= −0.439, p < 0.001), diffi culties with impulse control (standardized β = −0.419, p < 0.001) and diffi culties engaging in goal-directed behaviours (standardized β = −0.387, p < 0.001). In addition, treatment condition predicted symptom severity: depression (standardized β = −0.523, p < 0.001), anxiety (stan- dardized β = −0.444, p < 0.001) and suicidal cognitions (standardized β = −0.410, p < 0.001). Thus, the second requirement for mediation was fulfi lled.

Table 2. Pearson correlations between the number of episodes of DSH at baseline and emotion regulation difﬁ cul- ties in c patients (n = 90)

DSH Aware Clarity Non-accept Strategies Impulses Goals

DSH – 0.29* 0.01 19 0.23* 0.10 0.17**

Lack of awareness of emotions – 0.35** 0.19 0.10 0.08 0.05

Lack of clarity of emotions – 0.34** 0.25* 0.53** 0.26*

Non-acceptance of emotions – 0.48** 0.40** 0.48**

Limited strategies – 0.53** 0.55**

Difﬁ culties impulse control – 0.72**

Difﬁ culties goal-directed behaviours –

* Correlation is signiﬁ cant at the 0.05 level (two-tailed).

** Correlation is signiﬁ cant at the 0.01 level (two-tailed).

DSH = deliberate self-harm.

Table 3. Pearson correlations between the number of episodes of DSH at baseline and measures of symptom severity (depression, anxiety and suicidal cognitions) in DSH patients (n = 90)

DSH Depression Anxiety Suicidal cognitions

DSH – 0.28** 0.26* 0.19

Depression (BDI-II) – 0.53** 0.74**

Anxiety (SCL-90) – 0.51**

Suicidal cognitions (SCS) –

** Correlation is signiﬁ cant at the 0.01 level (two-tailed).

BDI-II = Beck Depression Inventory II.

(9)

Relationship between Change in the Proposed Mediators and Change in Outcome (Mediation Test—Step 3)

The analyses of Step 3 were performed with the data of the patients who had received the CBT intervention. The data of the patients who received TAU only were not included in these analyses. To determine whether the residualized change scores of the mediators correlated with the residualized change score of DSH (Path B), we used separate linear regression analyses for each of the proposed mediators. Change in diffi culties with impulse control predicted change in DSH (standardized β = 0.285, p = 0.047). Change in diffi culties engaging in goal-directed behaviours also predicted change in DSH (standardized β = 0.319, p = 0.026). However, the other emotion regulation diffi culties subscales did not predict change in DSH: lack of awareness (standardized β = 0.206, p = 0.157), lack of clarity (standardized β = 0.100, p = 0.492), non-acceptance of emotions (standardized β = 0.231, p = 0.111) and lack of strategies (standardized β = 0.207, p = 0.153).

Furthermore, the residualized change scores of the measurements of symptom severity did not predict change in DSH either: depression (standardized β

= 0.018, p = 0.900), anxiety (standardized β = −0.006, p = 0.966) and suicide cognitions (standardized β = 0.117, p = 0.423). Thus, the third requirement for mediation was only fulfi lled for diffi culties controlling impulses and for diffi culties engaging in goal-directed behaviours.

Effects of Treatment on DSH after Controlling for the Effects of the Proposed Mediators (Mediation Test—Step 4)

To assess whether the predictive effects of condition on the outcome DSH (Path C) were signiﬁ cantly reduced when the mediator was

statistically controlled (Path C′), we used separate regression analyses for difﬁ culties with impulse control and difﬁ culties engaging in goal-directed behaviours.

The residualized change score of DSH was entered as the dependent variable. The residualized change scores of the proposed mediators (diffi culties with impulse control or diffi culties engaging in goal-directed behaviours) and treatment condition (coded 0 for TAU and 1 for CBT) were entered as predictor variables. Results are presented in Figure 2. Results indicated that the change in DSH was mediated by the change in diffi culties with impulse control; condition was no longer a signifi cant predictor of change in DSH (standardized β = −0.208, p = 0.06) whereas diffi culties with impulse control remained signifi cant (standardized β = −0.216, p = 0.05). The entire mediation model accounted for Table 4. Pearson correlations between the number of episodes of DSH at 9-month follow-up and emotion regulation diffi culties in DSH patients (n = 90)

DSH Aware Clarity Non-accept. Strategies Impulses Goals

DSH – 0.30* 0.13 0.31** 0.31** 0.27* 0.30**

Lack of awareness of emotions – 0.62** 0.54** 0.57** 0.47** 0.47**

Lack of clarity of emotions – 0.62** 0.64** 0.71** 0.58*

Non-acceptance of emotions – 0.81** 0.76** 0.75**

Limited strategies – 0.81** 0.82**

Difﬁ culties impulse control – 0.86**

Difﬁ culties goal-directed behaviours –

* = p<.05; ** = p<.01; *** = p<.001

Condition DSHΔ

ImpulsesΔ GoalsΔ

DSHΔ

Condition Condition DSHΔ

C' =-.208 C' =-.196

B = .216* A = -.387*** B = .264*

C = -.298**

Condition DSHΔ

C = -.298**

A = -.419***

Figure 2. Intent to treat sample: N = 90. Path models of mediated intervention effects with deliberate self-harm (DSH) and two of the Diffi culties in Emotion Regula- tion Questionnaire subscales: diffi culties with impulse control and diffi culties with goal-directed behaviours.

Dotted lines indicate that the mediator variable was controlled for. ∆ = change

(10)

12.8% of the change in DSH. However, the difference between C (standardized β = −0.298) and C′ (standardized β = −0.208) is small (see Figure 2). The results also indicated that change in DSH was mediated by change in difﬁ culties engaging in goal-directed behaviours; condition was no longer a signiﬁ cant predictor of DSH (standardized β

= −0.196, p = 0.07) whereas diffi culties engaging in goal-directed behaviours remained signifi cant (standardized β = 0.264, p = 0.02). The entire media- tion model accounted for 14.9% of the change in DSH. As for diffi culties with impulses, the difference between C (standardized β = −0.298) and C′

(standardized β = −0.196) for difﬁ culties engaging in goal-directed behaviours is small (see Figure 2). The above analyses are based on the intent-to- treat sample (n = 90). Analyses with the completers sample (n = 73) led to similar results for each of the four steps of Baron and Kenny (1986).

DISCUSSION

Cognitive-behavioural therapy for DSH seems to be an effective intervention for DSH as shown in previous studies (Brown et al., 2005; Palmer et al., 2006; Slee et al., 2008a). Despite the demonstrated efﬁ cacy of CBT for DSH, very little is known about the mediating variables that lead to the reduction of DSH. What mediates treatment in this time-limited CBT? The theoretical model of DSH predicts that emotion regulation difﬁ culties are important mediators of treatment change (Slee et al., 2007).

Consistent with the mediation hypothesis, our fi ndings show that change in diffi culties with impulse control and change in diffi culties engaging in goal- directed behaviours partially mediate reduction in DSH. However, although the observed effects for- mally satisfy the Baron and Kenny conditions for partial mediation, the effects of the mediators are small. This suggests that the mediators are not a suffi cient or necessary condition for the treatment effect to occur. Furthermore, only two of the four emotion regulation variables that were expected to mediate change were found to mediate treatment effect. Clearly, there is a need for further testing.

Nevertheless, reduction of impulse control diffi - culties and diffi culties engaging in goal-directed behaviours (partially) mediated change. These fi ndings seem to suggest that these specifi c aspects of emotion regulation could be relevant targets of treatment. The process of CBT might involve the modulation of emotional arousal, reducing impul- sivity and facilitating behaviour that is in accordance with desired goals (Gratz & Roemer, 2004).

Although the CBT intervention consisted of a number of different therapeutic elements to target emotion regulation diffi culties, only diffi - culties with impulse control and diffi culties with goal-directed behaviours were found to partially mediate treatment effect. These fi ndings might suggest that the primary focus of the CBT had been on the ability to control behaviour in the context of emotional distress, and not so much on the awareness and acceptance of emotions. The high rate of repetition of self-harm found in the present study sample might have asked for greater use of interventions focusing on behavioural control. Aware- ness and acceptance of emotions might constitute a worthwhile treatment focus when patients are no longer in crisis. For example, Mindfulness- Based Cognitive Therapy is aimed at relapse prevention of DSH when patients are no longer acutely depressed (Williams, Duggan, Crane,

& Fennell, 2006). Future studies could compare an intervention similar to the CBT intervention of the present study with a mindfulness-based intervention in order to investigate whether the effectiveness of CBT and mindfulness-based interventions can be attributed to different underlying mechanisms.

Besides these fi ndings on specifi c emotion-regulation diffi culties, there are some other results that deserve attention. Although the CBT intervention signifi cantly reduced depression, anxiety and suicidal cognitions, these measures of symptom severity did not play a mediating role. Before the introduction of DSH-specifi c treatments such as dialectical behaviour therapy (Linehan, 1993;

Miller, Rathus, & Linehan, 2007) and cognitive therapy for suicidal patients (Berk et al., 2004), DSH was treated indirectly by treating associated mental disorders, such as depression (Miller et al., 2007). Since the CBT intervention in the present study focused on emotion regulation difﬁ culties associated with repetition of DSH and not on depressive or other disorders, the results of this study seem to indicate that a treatment focusing on emotion regulation difﬁ culties can be used success- fully to reduce symptom severity as well as DSH.

These fi ndings are consistent with DSH-specifi c therapies, suggesting that reduction of DSH can best be established by addressing emotion regulation diffi culties.

However, when interpreting these results, it must be taken into account that six variables represent- ing emotion regulation and only three variables assessing symptom severity were tested. There- fore, the likelihood of chance positive ﬁ ndings was

(11)

much higher for the ﬁ rst group of variables than for the second. Future studies might include an equal number of each group of variables to rule out the possibility of chance positive ﬁ ndings.

According to the theoretical ideas presented in the introduction, DSH is due to defi cits in emotion regulation. Surprisingly, however, the correlations reported in Table 2 show that only half of the emotion regulation variables assessed were signifi - cantly related to DSH at baseline and correlations were rather small. Interestingly, ‘diffi culties with impulse control’, one of the variables found to partially meet criteria for mediation, did not show a signifi cant correlation with DSH at all. This raises the question whether this variable can be regarded as a mediator for the reduction in DSH if it is not related to DSH at baseline. The lack of signifi - cant correlations is a challenge to the underlying theoretical model. The results might suggest that emotion regulation diffi culties are correlated with DSH in the general population or general clinical samples, but do not specifi cally predict the severity of DSH in samples that all show high levels of DSH. The fact that there are signifi cant correlations between DSH and emotion regulation diffi - culties post-treatment may be consistent with this view, as there is a greater variance in DSH at this stage.

It is also important to discuss the possible role of third variables. There are a number of variables that might explain some of the treatment effects, such as DSM-IV axis II diagnoses and dose and type of treatment received as part of TAU. However, in the present study, the presence of personality disorders was not assessed with a structured clinical interview. Therefore, personality disorders can not be ruled out as a third variable. With regard to dose and type of treatment received as part of TAU, previous analysis showed that both conditions received a comparable level of psychotherapy, psychotropic medication and psychiatric hospitalizations in TAU (see Slee et al., 2008a). However, we did not record speciﬁ c types of psychotherapy or psychotropic medication in TAU. Therefore, it is unclear if the conditions were equivalent in this respect and whether TAU has inﬂ uenced the mediator analyses.

Given the limitations of the present study, there is a need for more studies on the same emotion regulation variables. If future studies will replicate these fi ndings and fi nd stronger evidence for specifi c mediators, it might be interesting to follow the advice of Kraemer, Wilson, Fairburn, and Agras (2002) and develop treatment enhanced in those

components associated with the mediators to see whether we can identify the active ingredients of the intervention (e.g., teaching emotion-regulation skills). The beneﬁ ts of uncovering mechanisms of change would be considerable.

Although this CBT intervention is based on the theoretical assumption that a reduction in DSH is the result of changes in emotion regulation dif- ﬁ culties, other mechanisms of change may also be at work. For instance, Michel and Valach (2001) suggest that a shared mechanism underlying effective treatments of DSH is a strong therapeutic alliance, which allows for a meaningful discourse about DSH, whereas Bateman and Fonagy (2004) point to the importance of enhancement of mentalization (i.e., the ability to understand and reﬂ ect upon one’s own and other’s internal states and their relationship to behaviours). Given the CBT’s emphasis on a strong therapeutic alliance as well as on understanding emotions, either of these may be mechanisms of change of the present CBT intervention. Future research may clarify the exact underlying mechanisms.

There are also some other limitations that deserve attention. First, it remains unclear whether the mediation effects would also have been found in individuals who were excluded from the study.

In addition, the study primarily involved young females of Dutch nationality and with a history of chronic DSH. This absence of diversity limits the generalizability of ﬁ ndings. Second, the study does not distinguish between suicidal acts and self- injury. Therefore, it is unclear if different mediators can be found for these behaviours. Third, session- by-session assessment of emotion regulation difﬁ culties might have given more insight in maintaining factors of DSH. In addition, more frequent assessments would have allowed us to perform a mediation analysis along the lines of Kraemer et al.

(2002). Finally, although the proposed mediators were all measured by reliable and valid self-report questionnaires, their assessment could have been improved by using experimental tests of emotion regulation in addition to self-reports.

In conclusion, a manualized time-limited CBT intervention for DSH changes difﬁ culties with goal- directed behaviours and difﬁ culties with impulse control, which partially mediate reduction in DSH.

Furthermore, since depression, anxiety and suicidal cognitions did not seem to play a mediating role, there is further evidence for DSH-speciﬁ c interventions aimed at decreasing emotion regulation dif- ﬁ culties. However, since the mediation effects are small, there is a need for further testing.

(12)

ACKNOWLEDGEMENTS

We are grateful to the patients who participated in this study. This study was supported by The Netherlands Organization for Health Research and Development (ZonMw) (contract grant number:

2100.0068).

REFERENCES

Arrindell, W.A., & Ettema, J.H.M. (1986). Dutch trans- lation of the SCL-90. Lisse, the Netherlands: Swets &

Zeitlinger.

Baron, R.M., & Kenny, D.A. (1986). The moderator- mediator variable distinction in social psychological research: Conceptual, strategic and statistical consid- erations. Journal of Personality and Social Psychology, 51, 1173–1182.

Bateman, A.W., & Fonagy, P. (2004). Psychotherapy for borderline personality disorder: Mentalization based treat- ment. Oxford: Oxford University Press.

Beck, A.T., Steer, R.A., & Brown, G.K. (1996a). Manual for the Beck Depression Inventory II. San Antonio, TX:

Psychological Corporation.

Beck, A.T., Steer, R.A., Ball, R., & Ranieri, W.F. (1996b).

Comparison of beck depression inventories IA and II in psychiatric outpatients. Journal of Personality, 67, 588–597.

Berk, M.S., Henriques, G.R., Warman, D.M., Brown, G.K.,

& Beck, A.T. (2004). A cognitive therapy intervention for suicide attempters: An overview of the treatment and case examples. Cognitive and Behavioral Practice, 11, 265–277.

Bijttebier, P., & Vertommen, H. (1999). Coping strate- gies in relation to personality disorders. Personality and Individual Differences, 26, 847–856.

Brown, G.K., Ten Have, T., Henriques, G.R., Xie, S.X., Hollander, J.E., & Beck, A.T. (2005). Cognitive therapy for the prevention of suicide attempts. A randomized controlled trial. Journal of the American Medical Associa- tion, 294, 563–570.

Chapman, A.L., Gratz, K.L., & Brown, M.Z. (2006).

Solving the puzzle of deliberate self-harm: The experi- ential avoidance model. Behaviour Research and Therapy, 44, 371–394.

Davidson, K., Norrie, J., Tyrer, P., Gumley, A., Tata, P., Murray, H., & Palmers, S. (2006). The effectiveness of cognitive behaviour therapy for borderline personal- ity disorder: Results from the BOSCOT trial. Journal of Personality Disorders, 20, 450–465.

De Leo, D., Bille-Brahe, U., Kerkhof, A., & Schmidtke, A.

(2004). Suicidal behavior. Gottingen, Germany: Hogrefe

& Huber.

Derogatis, L.R., Lipman, R.S., & Covi, L. (1973). The SCL- 90: An outpatient psychiatric rating scale. Psychophar- macology Bulletin, 19, 13–28.

Ebner-Priemer, U.W., Welch, S.S., Grossman, P., Reisch, T., Linehan, M.M., & Bohus, M. (2007a). Psychophysi- ological ambulatory assessment of affective dysregu- lation in borderline personality disorder. Psychiatry Research, 150, 265–275.

Ebner-Priemer, U.W., Kuo, J., Kleindienst, N., Welch, S.S., Reisch, T., Reinhard, I., Lieb, K., Linehan, M.M., Bohus, M. (2007b). State affective instability in borderline personality disorder assessed by ambulatory monitoring. Psychological Medicine, 37, 961–970.

Giesen-Bloo, J., van Dyck, R., Spinhoven, P., van Tilburg, W., Dirksen, C., van Asselt, T., Kremers, I., Nadort, M., Arntz, A. (2006). Outpatient psychotherapy for borderline personality disorder: A randomized trial of schema focused therapy versus transference focused psycho- therapy. Archives of General Psychiatry, 63, 649–658.

Gratz, K.L., & Roemer, L. (2004). Multidimensional assessment of emotion regulation and dysregulation:

development, factor structure, and initial validation of the difﬁ culties in emotion regulation scale. Journal of Psychopathology and Behavioural Assessment, 26, 41–

54.

Gratz, K.L., Rosenthal, M.Z., Tull, M.T., Lejuez, C.W., &

Gunderson, J.G. (2006). An experimental investigation of emotion dysregulation in borderline personality disorder. Journal of Abnormal Psychology, 115, 850–855.

Hawton, K., Zahl, D., & Weatherall, R. (2003). Suicide following deliberate self-harm: Long-term follow-up patients who presented to a general hospital. British Journal of Psychiatry, 182, 537–542.

Kamphuis, J.H., Ruyling, S.B., & Reijntjes, A.H. (2007).

Testing the emotion regulation hypothesis among self-injuring females. Evidence for differences across mood states. Journal of Nervous and Mental Disorders, 195, 912–918.

Kraemer, H.C., Wilson, T., Fairburn, C.G., & Agras, W.S.

(2002). Mediators and moderators of treatment effects in randomized clinical trials. Archives of General Psy- chiatry, 59, 877–883.

Leible, T.L., & Snell, W.E., Jr. (2004). Borderline personality disorder and multiple aspects of emotional intelligence. Personality and Individual Differences, 37, 393–404.

Levine, D., Marziali, E., & Hood, J. (1997). Emotion processing in borderline personality disorders. Journal of Nervous and Mental Diseases, 185, 240–246.

Linehan, M.M. (1993). Cognitive-behavioral treatment of borderline personality disorder. New York: The Guilford Press.

Linehan, M.M., Comtois, K.A., Murray, A.M., Brown, M.Z., Gallop, R.J., Heard, H.L., Korslund, K.E., Tutek, D.A., Reynolds, S.K., Lindenboim, N. (2006). Two-year randomized controlled trial and follow-up of dialectical behavior therapy vs therapy by experts for suicidal behaviors and borderline personality disorder.

Archives of General Psychiatry, 63, 757–766.

Michel, K., & Valach, L. (2001). Suicide as goal-directed action. In K. van Heeringen (Ed.), Understanding sui- cidal behaviour: The suicidal process approach to research and treatment (pp. 230–254) Chichester: Wiley & Sons.

Miller, A.L., Rathus, J.H., & Linehan, M.M. (2007). Dialec- tical behavior therapy with suicidal adolescents. New York:

The Guilford Press.

Palmer, S., Davidson, K., Tyrer, P., Gumley, A., Tata, P., Norrie, J., Murray, H., & Seivewright, H. (2006).

The cost-effectiveness of cognitive behavior therapy for borderline personality disorder: results from the

(13)

BOSCOT trial. Journal of Personality Disorders, 20, 466–

481.

Rudd, M.D., Joiner, T., & Rajab, M.H. (2001). Treating sui- cidal behaviour. An effective, time-limited approach. New York: The Guilford Press.

Schmidtke, A., Bille-Brache, U., De Leo, D., & Kerkhof, A. (2004). Suicidal behaviour in Europe. Gottingen, Germany: Hogrefe & Huber.

Slee, N., Arensman, E., Garnefski, N., & Spinhoven, Ph.

(2007). Cognitive behavioral therapy for Deliberate Self-Harm. Crisis, 28, 175–182.

Slee, N., Garnefski, N., Van der Leeden, R., Arensman, E.,

& Spinhoven, Ph. (2008a). Outcomes of a randomized controlled trial of a cognitive behavioral intervention for Deliberate Self-Harm patients. British Journal of Psy- chiatry, 192, 202–211.

Slee, N., Garnefski, N., Spinhoven, Ph., & Arensman, E. (2008b). The inﬂ uence of cognitive emotion regulation strategies and depression severity on Deliberate Self-Harm. Suicide and Life-Threatening Behavior, (in press)

Stone, A.A., Broderick, J.E., Shiffman, S.S., & Schwartz, J.E. (2004). Understanding recall of weekly pain from a momentary assessment perspective: absolute agreement, between- and within-person consistency, and judged change in weekly pain. Pain, 107, 61–

69.

Tyrer, P., Thompson, S., Schmidt, U., Jones, M., Knapp, M., Davidson, J., Knapp, M., MacLeod, A.,

& Catalan, J. (2003). Randomized controlled trial of brief cognitive behaviour therapy versus treatment as usual in recurrent deliberate self-harm: the POPMACT study. Psychological Medicine, 33, 969–

976.

Williams, J.M.G., Duggan, D.S., Crane, C., & Fennell, M.J.V. (2006). Mindfulness-based cognitive therapy for prevention of recurrence of suicidal behavior. Journal of Clinical Psychology, 62, 201–210.

Yen, S., Zlotnick, C., & Costello, E. (2002). Affect regulation in women with borderline personality disor- der traits. Journal of Nervous and Mental Disease, 190, 693–696.